Siddhisanket (Sid) Raskar

I am a Assistant Computer Scientist in the AI/ML group at Argonne National Laboratory. I work on AI architectures for scientific machine learning and on the design of next-generation AI architectures for science. My recent work involves performance benchmarking, modeling, and optimizations targeting modern ML workloads (e.g LLM, LMM, CNN, GNN, etc.) for heterogeneous and parallel architectures (e.g., GPUs and AI accelerators).

I obtained my Ph.D. under the guidance of Guang R. Gao on the topic of "Dataflow Software Pipelining” at the University of Delaware. I also have a Master’s in Computer Science from the University of Delaware and a Bachelor’s in Computer Engineering from the University of Pune, India.

My research interesrts are at the intersection of Machine Learning, High Performance Computing and Computer Architectures.

Research Interests :

  • Machine Learning and AI Architectures
  • High-Performance Computing
  • Computing and Dataflow Models
  • Computer Architecture and Systems
  • Compilers Technology and Runtime Systems


Professional Service


Selected Talks


Publications

  • [PMBS'24] LLM-Inference-Bench: Inference Benchmarking of Large Language Models on AI Accelerators
    Krishna Teja Chitty-Venkata, Siddhisanket Raskar, Bharat Kale, Farah Ferdaus, Aditya Tanikanti, Ken Raffenetti, Valerie Taylor, Murali Emani, Venkatram Vishwanath
    2024 IEEE/ACM International Workshop on Performance Modeling, Benchmarking and Simulation of High-Performance Computer Systems (PMBS), Atlanta, GA, USA, 2024.
    Paper
  • [HCW'24] Toward a Holistic Performance Evaluation of Large Language Models Across Diverse AI Accelerators
    Murali Emani, Sam Foreman, Varuni Sastry, Zhen Xie, Siddhisanket Raskar, et al.
    2024 IEEE International Parallel and Distributed Processing Symposium Workshops (IPDPSW), San Francisco, CA, USA, 2024.
    Paper
  • [SIGMETRICS'24] Thorough Characterization and Analysis of Large Transformer Model Training At-Scale
    Scott Cheng, Jun-Liang Lin, Murali Emani, Siddhisanket Raskar, Sam Foreman, Zhen Xie, Venkatram Vishwanath, Mahmut Kandemir
    ACM Sigmetrics/IFIP Performance 2024.
    Paper
  • [AppliedSciences'24] Cross-Feature Transfer Learning for Efficient Tensor Program Generation
    Verma, Gaurav, Siddhisanket Raskar, Murali Emani, and Barbara Chapman
    Applied Sciences 14, no. 2: 513
    Paper
  • [WACCPD'23] Performance of Triangle Counting on Graphcore's IPU Architecture
    Reet Barik, Siddhisanket Raskar, Murali Emani, Venkatram Vishwanath
    Tenth Workshop on Accelerator Programming and Directives (WACCPD 2023) at Supercompututing 2023, November 12-17, Denver, Colorado, USA.
    Paper
  • [CARLA'23] Towards Fault Tolerance and Resilience in the Sequential Codelet Model
    Diego A Roa Perdomo, Rafael A Herrera Guaitero, Dawson Fox, Hervé Yviquel, Siddhisanket Raskar, Xiaoming Li, Jose M Monsalve Diaz
    The Latin America High Performance Computing Conference (CARLA2023), September 18-22, 2023, Cartagena de Indias, Colombia.
    Paper
  • [ArXiv'23] A Comprehensive Performance Study of Large Language Models on Novel AI Accelerators
    Murali Emani, Sam Foreman, Varuni Sastry, Zhen Xie, Siddhisanket Raskar, William Arnold, Rajeev Thakur, Venkatram Vishwanath, Michael E Papka
    Paper
  • [LDRD'23] LLVM's Frontend and Runtime modifications to support OpenMP in the GraphCore architecture
    Jose M Monsalve Diaz, Rodrigo Ceccato de Freitas, Esteban Rangel, Siddhisanket Raskar
    Report
  • [EuroPar'23] TrainBF: High-Performance DNN Training Engine using BFloat16 on AI Accelerators
    Zhen Xie, Siddhisanket Raskar, Murali Emani, Venkatram Vishwanath
    29th International European Conference on Parallel and Distributed Computing 2023 (EuroPar'23), Limassol, Cyprus.
    Paper
  • [ICPE'23] Implementation of Dataflow Software Pipelining for Codelet Model
    Siddhisanket Raskar, Jose M Monsalve Diaz, Thomas Applencourt, Kalyan Kumaran, and Guang Gao
    Proceedings of the 2023 ACM/SPEC International Conference on Performance Engineering (ICPE’23), 2023, Coimbra, Portugal.
    Paper
  • [ExHET'23] Transfer Learning Across Heterogeneous Features For Efficient Tensor Program Generation
    Gaurav Verma, Siddhisanket Raskar, Zhen Xie, Abid M Malik, Murali Emani, Barbara Chapman
    Proceedings of the 2nd International Workshop on Extreme Heterogeneity Solutions (ExHet)
    Paper
  • [PMAM'23] Towards Maximum Throughput of Dataflow Software Pipeline under Resource Constraints
    Siddhisanket Raskar, Thomas Applencourt, Kalyan Kumaran, and Guang Gao
    International Workshop on Programming Models and Applications for Multicores and Manycores (PMAM ’23), 2023, Montreal, QC, Canada.
    Paper
  • [PMBS'22] A Comprehensive Evaluation of Novel AI Accelerators for Deep Learning Workloads
    Murali Emani, Zhen Xie, Siddhisanket Raskar et al.
    2022 IEEE/ACM International Workshop on Performance Modeling, Benchmarking and Simulation of High Performance Computer Systems (PMBS), Dallas, TX, USA.
    Paper
  • [LDRD'22] A pathway to OpenMP In the GraphCore Architecture
    Jose M Monsalve Diaz, Esteban Rangel, Siddhisanket Raskar, Johannes Doerfert
    Report
  • [ScaDL'22] Throughput-oriented and Accuracy-aware DNN Training with BFloat16 on GPU
    Zhen Xie, Siddhisanket Raskar, and Murali Emani
    2022 IEEE International Parallel and Distributed Processing Symposium Workshops (IPDPSW)
    Paper
  • [PHD] Dataflow Software Pipelining for Codelet Model using Hardware-Software Co-design
    Siddhisanket Raskar
    University of Delaware, Summer 2021.
    Thesis
  • [CAPSL'21] clCodeletPipe API Documentation: Implementation of Codelet Pipe on Intel Iris Pro Architecture
    Siddhisanket Raskar, Thomas Applencourt, Kalyan Kumaran, and Guang Gao
    CAPSL Technical Memo 140, May 2021.
    Paper
  • [CAPSL'21] Realization of Dataflow Software Pipelining for Codelet Model using Hardware-Software Co-design
    Siddhisanket Raskar, Thomas Applencourt, Kalyan Kumaran, and Guang Gao
    CAPSL Technical Memo 139, May 2021.
    Paper
  • [IPDRM'20] CODIR: Towards an MLIR Codelet Model Dialect
    Ryan Kabrick, Diego A. Roa Perdomo, Siddhisanket Raskar and Jose Monsalve Diaz and Dawson Fox and Guang R. Gao
    2020 IEEE/ACM Fourth Annual Workshop on Emerging Parallel and Distributed Runtime Systems and Middleware (IPDRM), GA, USA, 2020
    Paper
  • [IPDRM'20] DEMAC: A Modular Platform for HW-SW Co-Design
    Diego A. Roa Perdomo, Ryan Kabrick, Jose MMonsalve Diaz, Siddhisanket Raskar, Dawson Fox and Guang R. Gao
    2020 IEEE/ACM Fourth Annual Workshop on Emerging Parallel and Distributed Runtime Systems and Middleware (IPDRM), GA, USA, 2020, pp. 25-32.
    Paper
  • [DFM'19] Position Paper: Extending Codelet Model for Dataflow Software Pipelining using Software-Hardware Co-Design
    Siddhisanket Raskar, Thomas Applencourt, Kalyan Kumaran, and Guang Gao
    2019 IEEE 43rd Annual Computer Software and Applications Conference (COMPSAC), Milwaukee, WI, USA, 2019.
    Paper
  • [DFM'19] Toward A High-Performance Emulation Platform for Brain-Inspired Intelligent Systems: Exploring Dataflow-Based Execution Model and Beyond
    Sihan Zeng, Jose Monsalve Diaz, Siddhisanket Raskar
    2019 IEEE 43rd Annual Computer Software and Applications Conference (COMPSAC), Milwaukee, WI, USA, 2019.
    Paper
  • [CAPSL'20] Study of Dataflow Software Pipelining under Codelet Model using Cannon's Algorithm
    Siddhisanket Raskar, Jose Monsalve Diaz, Thomas Applencourt, Kalyan Kumaran, Guang Gao
    CAPSL Technical Memo 135, February 2020.
    Technical Memo
  • [CAPSL'19] Brain-Flow: A brain inspired dataflow implementation using DEMAC
    Diego Roa, Ryan Kabrick, Siddhisanket Raskar, Jose Monsalve Diaz, Guang Gao
    CAPSL Technical Memo 134, October 2019.
    Technical Memo
  • [CAPSL'19] Extending Codelet Model for Dataflow Software Pipelining using Software-Hardware Co-design
    Siddhisanket Raskar, Thomas Applencourt, Kalyan Kumaran, and Guang Gao
    CAPSL Technical Memo 133, June 2019.
    Technical Memo
  • [ETI'15] Legacy MPI Codes and its interoperability with fine grain task-parallel runtime systems for Exascale
    Siddhisanket Raskar and Guang Gao
    ETI Technical Report 4, August 2015. (Available on Request)